noise floor

[Beth Hutchinson]

Hi Gary,
Thanks for distributing/supporting NODDI so nicely.  I wonder if you could tell me how the noise floor is handled by the NODDI Matlab toolbox.
In particular, is it included in the model?  I notice this in the text of your 2012 paper:
"The procedure determines the maximum likelihood estimates of the parameters, using a Rician noise model,   with   the   Gauss-Newton   nonlinear   optimization   technique.
 The starting point for the nonlinear optimization is obtained from an initial brute-force search over a coarse and regular grid of physically plausible settings."
and I notice in your matlab fitting code a GridSearch algorithm with a Rician distribution, but I am failing to connect the two conceptually to know how the noise floor is/isn't modeled when I use this implementation on my own data (which is ex-vivo modeling). 
Thanks much far any insight on this.
Best,
Beth

[Hui Zhang]

Hi Beth,

Noise floor is a very general concept that can arise from a range of noise models.  One example is the Rician noise model; another is the Chi-squared noise model.  So the current implementation assumes the noise can be modelled as Rician.  If this assumption is reasonable, it can therefore handle the noise floor resulting from this type of noise.

Gary

--
You received this message because you are subscribed to the Google Groups "NODDI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to noddi+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ajay Kurani]

Hi Gary,

   If we have denoising as part of our pre-processing pipeline is there a way to disable this option of using a Rician noise model?  

Are there any steps (aside from GridSearchRician) that uses these parameters?

Thanks,

Ajay 

[Hui Zhang]

Hi Ajay,

When the noise level is very low, the Rician noise model becomes the Gaussian noise model.  You don't have to anything special.

Gary